highly enhanced and stable activity of defect-induced...
TRANSCRIPT
Materials Today d Volume 20, Number 9 d November 2017 RESEARCH
Highly enhanced and stable
Resea
rch
activity of defect-induced titania
EARCH:Original
nanoparticles for solar light-driven
CO2 reduction into CH4 RES
Saurav Sorcar 1, Yunju Hwang 1, Craig A. Grimes 2, Su-Il In 1,⇑
1 Department of Energy Science and Engineering, DGIST, 333 Techno Junga
ng-daero, Hyeonpung-myeon, Dalseong-gun, Daegu 42988, Republic of Korea2 Flux Photon Corporation, 116 Donmoor Court, Garner, NC 27529, United StatesPhotocatalytic reduction of CO2 to fuel offers an exciting opportunity for helping to solve currentenergy and global warming problems. Although a number of solar active catalysts have been reported,most of them suffer from low product yield, instability, and low quantum efficiency. Therefore, thedesign and fabrication of highly active photocatalysts remains an unmet challenge. In the currentwork we utilize hydrogen-doped, blue-colored reduced titania for photocatalytic conversion ofCO2 into methane (CH4). The photocatalyst is obtained by exposure of TiO2 to NaBH4 at 350 �C for0.5 h. Sensitized with Pt nanoparticles, the material promotes solar spectrum photoconversion of CO2
to CH4 with an apparent quantum yield of 12.40% and a time normalized CH4 generation rate of80.35 lmol g�1 h�1, which to the best of our knowledge is a record for photocatalytic-based CO2
reduction. The material appears intrinsically stable, with no loss in sample performance over five 6 hcycles, with the sample heated in vacuum after each cycle.
IntroductionAs is now well known, anthropogenic emission of greenhousegases, particularly CO2, is a significant factor driving global cli-mate change [1]. Sustainable, low carbon, readily portable fuelsare one of the most pressing needs of modern society. To thatend, a number of studies have investigated conversion of CO2
into products such as CH4 [2], CH3OH [3], and CO [4]. For thispurpose, a variety of semiconductors have been studied, to citebut a few examples ZnGa2O4 [5], CdS [6], TiO2 [6,7], and Ru/RuOx loaded TiO2 [8,9]. However, despite the many efforts, thephotocatalysts studied to date suffer from low photoconversionefficiency and limited stability.
The surface structure of a photocatalyst plays a crucial role indetermining photocatalytic activity. A recent report [10] suggeststhat creating oxygen vacancies (Vo) and Ti3+ states in titaniaresults in an upward shift of the valence band maximum(VBM) and a downward shift of the conduction band maximum(CBM), in turn enhancing light absorption properties. In initial
⇑ Corresponding author.
E-mail address: In, S.-I. ([email protected]).
1369-7021/� 2017 Elsevier Ltd. All rights reserved. https://doi.org/10.1016/j.mattod.2017.09.005
studies reduced titania has shown promising photocatalyticactivity for H2 generation by water photoelectrolysis [11] andCO2 photoreduction [12]. Liu et al. studied the role of surfaceVo and Cu species over reduced titania [4], and Zhu et al. synthe-sized Cu(I) supported titania nanosheets with defective {001}facets [10]. Pt, with a Fermi level (5.35–5.63 eV) more positivethan that of titania (4.67–4.70 eV), might be expected to effec-tively promote CO2 reduction [13]. However there are few reportson the photocatalytic performance of reduced titania sensitizedwith Pt nanoparticles, known to promote water-splitting ande�–h+ pair separation [12], in application to photocatalytic con-version of CO2 to hydrocarbon fuels.
Herein, we report a facile low-temperature synthesis route forfabricating reduced titania nanoparticles blue in color. The pho-tocatalyst, sensitized with Pt nanoparticles, is found to activelypromote CO2 photoreduction under solar spectrum light; withoptimized samples we obtain a maximum CH4 yield of 80.35lmol g�1 h�1 and an apparent quantum yield (AQY) of 12.40%;see the Appendix for details of the AQY calculation. While vari-ations on illumination (intensity, spectrum, duration) and sam-ple conditions (planar or bulk samples, chamber size, static ordynamic conditions) make direct comparison difficult, it
507
RESEA
RCH:O
riginal
Research
RESEARCH Materials Today d Volume 20, Number 9 d November 2017
appears, to the best of our knowledge, that this is the highest rateof CO2 to CH4 solar spectrum photocatalytic conversion, andAQY, achieved to date.
ExperimentalSynthesis of Blue Titania (BT)200-mg P25 procured from Degussa, was ground with variablequantities of Sodium Borohydride (NaBH4, 98%) procured fromAlfa Aesar using a mortar and pestle, then placed in a quartz tubefurnace and heated at 350 �C for 30 min under Ar flow. Afterannealing, the sample was washed with copious amounts ofdeionized (DI) water and then ethanol, a cycle repeated fivetimes, and then dried at 90 �C in a vacuum oven for 12 h. Differ-ent reduced titania samples were prepared by varying the NaBH4
amount (X), the samples, which are generally blue in color,denoted as BT-X, where X = 20, 30, 40 and 50 represents theamount of NaBH4 used in milligrams, mg.
Photodeposition of PtPt nanoparticles were photochemically deposited onto BT-30samples in the following manner: 100 mg of sample BT-30 wasadded to 20-ml DI water and 5 ml CH3OH (Duksan reagent).Variable concentrations of H2PtCl6 (HPLC grade, Sigma Aldrich)were added to the above mixture and stirred for 1 h under dark-ness in a closed system. The suspension was then irradiated witha 300W Xe lamp (Newport) the light intensity of which wasadjusted to 1 Sun (Air Mass (AM 1.5)) using a 1 Sun detector(Newport). The irradiation was performed for 2 h under mild stir-ring and the samples were then washed repeatedly with DI water,and finally dried at 90 �C in a vacuum oven for 12 h. The Pt pho-todeposited BT samples were named as Y-BT-30 (where Y = 0.25,0.30, 0.32, 0.35, 0.42 and 0.50 corresponding to theoretically cal-culated wt. % Pt). In an identical manner 0.35 wt. % Pt was pho-todeposited upon comparative control samples. A control samplein which pure P25 was treated in Ar atmosphere, without NaBH4
exposure, is identified as 0.35-P25 (Ar treated).
Photocatalyst characterizationX-ray diffraction (XRD) spectroscopy was performed on a Pana-lytical, Empyrean X-ray diffractometer using Cu kk radiation (k= 1.54 Å) operating at 40 kV and 30 mA. Raman analysis was car-ried out on a Raman Spectrometer, Nicolet Almeca XR, manufac-tured by Thermo Scientific. A 531 nm laser was used forexcitation. The lattice structure of photocatalyst was visualizedby a field emission transmission electron microscopy (FE-TEM)taken from Hitachi HF-3300 operating at 300 kV. The sampleswere dispensed upon Ni mesh grid with concentration of 0.05mg/ml prepared in ethanol and allowed to dry overnight. ImageJsoftware was used to estimate the Pt nanoparticle size distribu-tion and average particle size (davg) from the high resolutiontransmission electron microscopy (HR-TEM) images. The ele-mental composition and elemental mapping of 0.35-BT-30 (rep-resentative BT-30 sample with optimized Pt loading) weremeasured using the energy dispersive spectroscopy (EDS)attached with Hitachi HF-3300 FE-TEM. UV–visible diffuse reflec-tance spectra (UV–vis DRS) were measured using a Cary seriesUV–visible near infrared spectrophotometer with a diffuse reflec-tance accessory. Photoluminescence emission (PL) spectra were
508
measured on a Cary Eclipse fluorescence spectrophotometer(kexc = 320 nm) of Agilent Technologies having a diffuse reflec-tance accessory. X-ray photoelectron spectroscopy (XPS) andvalence band X-ray photoelectron spectroscopy (VB-XPS) weremeasured using Thermo VG, K-alpha using Al Kk line (148606eV) as the X-ray source. Electron paramagnetic resonance (EPR)spectra were recorded using a Jeol JES-FA100 spectrometer at100 K. The N2 sorption isotherms were measured at �196 �Con a BELSORP-miniII apparatus, with samples extensivelydegassed at 100 �C prior to the experiments. Surface areas werecalculated by the Brunauer–Emmett–Teller (BET) equation. Thepore volume was determined from the amount of N2 adsorbedat the highest relative pressure of (P/P0) = 0.99. Proton nuclearmagnetic resonance (1H NMR) measurement was recorded usinga Bruker Avance 400 spectrometer.
Photocatalytic CO2 reductionPhotocatalytic CO2 reduction experiments were carried out usingthe experimental setup shown in Fig. 1. In brief, a 40 mg sampleof the photocatalyst was evenly dispersed upon a porous frittedfilter disc that was placed at the center of the photoreactor. MoistCO2 gas was then continuously passed through the photoreactorat a flow rate of 40 ml/min. After 1 h purging the flow rate wasdecreased to 1.0 ml/min, and this rate was maintained for theentire photoreaction process. A 100 W Xe solar simulator (Oriel,LCS-100) with an AM1.5 filter was the light source. The concen-tration of the effluent gas, as a function of irradiation time, wasanalyzed every 30 min by a gas chromatograph unit (Shimadzu,GC-2014) equipped with an automated gas sampling valve,helium carrier gas, to determine CH4 product yield (as shownin Eq. (1)) [14] from the photocatalytic system. The GC wasequipped with a flame ionization detector (FID, Restek-Rt-Q-bond column, ID = 0.53 mm and length = 30 m). To evaluatephotocatalyst stability, the photocatalyst was repeatedly testedfor CO2 photoreduction; after each 6 h test period the photocat-alyst was vacuum annealed at 100 �C for 2 h, a process adoptedfrom previous reports for regenerating used photocatalyst[15,16].
Total CH4 yieldðlmolÞ
¼ ðCfinal;CH4 � Cinitial;CH4 Þ � volumetric flow of product gasAmount of photocatalyst used ðgÞ ð1Þ
Control and 13CO2 isotopic experiments were performed toinvestigate the origin of carbon involved in hydrocarbon gener-ation. The control test was carried out in (i) Ar/H2O(g) mixtureinstead of moist CO2 with similar experimental conditions, and(ii) a blank reactor test without any photocatalyst. For isotopicexperiments, 13CO2 (13C 99%) was purchased from Aldrich anddiluted in pure He gas (99.999%) to give 13CO2 with final a con-centration of 1000 ppm in He. The 13CH4 produced by moist13CO2 gas (13CO2 + H2O mixture) was analyzed using Gas chro-matography–Mass spectrometer (GC–MS) manufactured by Shi-madzu, GC–MS-QP2010 ultra (Restek Rt Q-bond column, ID =0.32 mm and length = 30 m). Moist 13CO2 was introduced intothe photocatalyst (40 mg)-loaded, AM 1.5G illuminated photore-actor and the gaseous products analyzed after 1 h, 2 h, and 3 h.
FIGURE 1
Setup used for the photocatalytic CO2 reduction into CH4 experiments, employing an online gas chromatography (GC) system for dynamic analysis ofgaseous products.
RES
EARCH:Original
Resea
rch
Materials Today d Volume 20, Number 9 d November 2017 RESEARCH
Results and discussionPowder X-ray diffraction patterns (XRD) of P25 [17] and blue tita-nia (BT)-X, where X represents the amount of NaBH4, in mg,used for reduction (0.5 h at 350 �C), are shown in Fig. 2a. TheXRD patterns indicate no significant appearance of any newphase and/or any structural change during the reduction process.As the amount of NaBH4 increases, the sample color changesfrom white, to light blue, to dark blue, suggesting defect forma-tion within the titania nanoparticles (Fig. 2b). Fig. S1 showsthe XRD patterns of various Pt-sensitized BT-30 samples. The sizeand content of the dispersed Pt nanoparticles is beyond the XRDdetection limit, in agreement with previous reports [18,19].
Fig. 3a shows a HR-TEM image of P25, where well-defined lat-tice fringes are seen with a 0.35-nm spacing confirming its crys-tallinity [12]; an image of sample BT-30, Fig. 3b, shows adisordered shell �1–2 nm thick and a crystalline core; the struc-tural difference is in accordance with Raman spectroscopy used
FIGURE 2
XRD patterns of: (a) different blue titania samples, and (b) depiction of colorchange for samples after reduction process with respect to amount ofNaBH4.
to confirm Vo and non-stoichiometry, see Fig. S2. BT-30, whichshowed optimal photocatalytic CO2 reduction activity, was sen-sitized with different amounts of co-catalyst Pt; such samples areidentified as Y-BT-30 where Y = 0.25, 0.30, 0.32, 0.35, 0.42 and0.50 corresponding to the theoretically calculated Pt wt. %.Fig. 3c exhibits a TEM image of sample 0.35-BT-30, in whichwell-dispersed Pt nanoparticles are evident; Fig. 3d is a magnifiedimage of the region enclosed by the red circle in Fig. 3c. In Fig. 3dwe see clear lattice fringes with interplanar distances of 0.35 nmand 0.23 nm, typical for the (101) plane of anatase [12,20] and(111) plane of face-centered cubic Pt [21]. Fig. 3e–h shows ele-mental mapping of sample 0.35-BT-30 with red, blue and greenspots respectively assigned to Ti, O and Pt. The Gaussian curve,Fig. 3j, indicates that the Pt nanoparticles are well dispersed withan average size of 2.42 ± 0.05 nm.
Textural properties of the samples were characterized by N2
adsorption–desorption isotherms with the results provided asFigs. S3 and S4 for blue titania and Pt deposited blue titania sam-ples, respectively. All BT samples possessed comparable BET sur-face areas of 52–56 m2 g�1, see Table 1. A reduction in BETsurface area of the Y-BT-30 samples was observed, indicating suc-cessful grafting of the Pt nanoparticles [21,22] see Fig. S4.
Fig. 4a displays UV–vis DRS spectra; all reduced titania sam-ples exhibit significant amounts of light absorption in the visibleregion. Tauc plots are used to calculate the bandgap, see Fig. S5and Table S1. The enhanced absorption is attributed to bothdisordered surface layers and trivalent titanium ions. It is knownthat hydrogen atoms in the amorphous layers can strongly inter-act with the Ti 3d and O 2p electrons, leading to a considerabledecrease in the band gap ascribed to mid-gap states formation[23,24]. Further, hydrogenation causes disorder among thesurface states inducing an upward shift of the valence bandmaximum [23,24], while Ti3+ and Vo reduce the conductionband minimum [23,24].
Fig. 4b compares the UV–vis DRS spectra of different Pt wt. %photodeposited BT-30 samples. The enhancement of visible lightabsorption with Pt sensitization is due to localized surface plas-mon resonance (LSPR) [18,25]. Fig. 4c displays the comparativephotoluminescence (PL) spectra of different reduced titania sam-ples; PL emission results from the recombination of free charge
509
FIGURE 3
HR-TEM images of: (a) P25, (b) BT-30 where the arrows indicate a disordered shell, (c) 0.35-BT-30, (d) enlarged image of 0.35-BT-30. Elemental mapping for: (e)0.35-BT-30, showing the regions of (f) Ti, (g) O, (h) Pt, (i) EDS, and (j) particle size distribution of the Pt nanoparticles in 0.35-BT-30.
TABLE 1
Physiochemical properties of photocatalysts.
Sample Pt size(nm)
Pt content(atomic %, from ICP-OES)
BET(m2 g�1)
Pore volume(cm3 g�1)
CH4 yield(lmol g�1 h�1)
AQY(%)
P25 – – 56.50 0.28 – –BT-20 – – 54.10 0.50 2.12 0.32BT-30 – – 55.70 0.49 2.79 0.41BT-40 – – 53.40 0.42 0.88 0.13BT-50 – – 52.10 0.40 0.68 0.100.25-BT-30 1.82 ± 0.07 0.18 49.3 0.38 10.16 1.520.30-BT-30 2.18 ± 0.04 0.22 46.4 0.35 20.03 3.080.32-BT-30 2.25 ± 0.02 0.28 45.9 0.34 48.96 7.530.35-BT-30 2.42 ± 0.05 0.33 45.5 0.32 80.35 12.400.42-BT-30 2.98 ± 0.06 0.38 44.7 0.31 66.68 10.320.50-BT-30 3.50 ± 0.18 0.42 43.9 0.30 41.13 6.34
RESEA
RCH:O
riginal
Research
RESEARCH Materials Today d Volume 20, Number 9 d November 2017
carriers [22,23], thus reduced emissions can be interpreted aslonger photogenerated charge lifetimes. Fig. 4d displays the PLspectra of different Pt wt. % sensitized BT-30 samples, with fur-ther quenching observed [13]. Electron paramagnetic resonance(EPR) and proton nuclear magnetic resonance spectroscopy (1HNMR) was carried out to confirm the Ti3+ states and H dopingin the blue titania samples, with the results and analysis givenin Fig. S6a and b, respectively.
XPS was used to examine the effect of hydrogenation on thechemical composition and oxidation states. Fig. 5a shows thesurvey scan spectra of BT-30, with a high resolution scan of theTi 2p region shown in Fig. 5b. For both samples two broad peakscentered at �458.28 eV and �464.08 eV are observed, corre-sponding to the characteristic Ti 2p3/2 and Ti 2p1/2 peaks ofTi4+ respectively [22,23]. In comparison to P25, the peaks of sam-ple BT-30 show a negative shift in binding energy, suggestingthat they have a different bonding environments [26]. The
510
O 1s region, Fig. 5c, shows a similar shift of binding energy forBT-30, which suggests a transfer of electrons to neighboring Vo
[27]. It should be noted that the generation of surface or subsur-face Vo accompanies the formation of Ti3+ for charge compensa-tion. XPS analysis was performed for all samples, see Fig. S7, andshows similar negative shifts in binding energy.
The density of states (DOS) were determined using valenceband X-ray photoelectron spectroscopy (VB-XPS) [26,27], seeFig. 5d. P25 shows the characteristic valence band DOS of titania,with the band edge at �2.29 eV (Fig. 4d). The VB-XPS of BT-30had an absorption onset located at 1.19 eV, whereas the maxi-mum energy associated with the band tail is blue-shifted to1.10 eV (Fig. 4d); similar changes in DOS have been previouslyreported for reduced titania [26,27]. Fig. S8 shows the VB-XPSresults and band energy diagrams for samples BT-20, BT-40,and BT-50; it is evident that the valence band tail increases withdegree of reduction. The XPS spectra of 0.35-BT-30 and other Pt
FIGURE 5
(a) Survey scan spectra of sample BT-30, (b) comparative core level XPS spectra of Ti 2p for BT-30 and P25 nanoparticles, (c) comparative core level XPSspectra of O 1s for BT-30 and P25 nanoparticles, and (d) comparative VB-XPS of BT-30 and P25 nanoparticles.
FIGURE 4
(a), (b) UV–vis DRS spectra, and (c), (d) PL spectra of different reduced titania samples, and different Pt wt. % photodeposited BT-30 samples, respectively.
RES
EARCH:Original
Resea
rch
Materials Today d Volume 20, Number 9 d November 2017 RESEARCH
511
RESEA
RCH:O
riginal
Research
RESEARCH Materials Today d Volume 20, Number 9 d November 2017
deposited reduced titania samples is given in Figs. S9 and S10respectively, confirming the presence of Pt.
Gas phase photocatalytic CO2 reduction results are shown inFig. 6. According to GC analysis CH4 was the predominanthydrocarbon product in all experiments, with vanishingly smallamounts of ethane. The CH4 evolution rate significantlyincreases with titania reduction, with a maximum CH4 evolutionrate of 2.79 lmol g�1 h�1 (8.39 lmol g�1) observed for sampleBT-30. The decrease in activity seen for BT-40 and BT-50 is pre-sumably due to excessive Ti3+, which can serve as recombinationcenters for photogenerated charges [11,15,23].
It is well known that noble metals create electron sinks as wellas active sites for photocatalytic CO2 reduction [18,28]. As evidentfrom the optical properties of the Y-BT-30 samples, Pt not onlyextends the photoresponse but also suppresses charge carrierrecombination by facilitating interfacial electron transfer. Pt load-ing values were experimentally verified by ICP-OES analysis, andfound to be quite close to theoretical calculations, Table 1.
The continuous CH4 production profile and rate of CH4 pro-duction for different Pt wt. % sensitized BT-30 under solar spec-trum illumination are shown in Fig. 6b and c, respectively.Among the studied samples, the rate of CH4 evolution reachesa maximum of 80.35 lmol g�1 h�1 [482.12 lmol g�1] for sample
FIGURE 6
Photocatalytic CO2 reduction results. Cumulative methane evolution under AMsamples. (c) Evolution amounts of CH4 in the first 6 h irradiation from photocatalBT-30; at the end of 2nd, 3rd and 4th cycle the photocatalyst is placed within
512
0.35-BT-30 (6 h) (Fig. 6c). A maximum apparent quantum yield(AQY) of 12.40%, see Table 1, is achieved. As far as we know,the CH4 production rate and AQY are the highest with regardsto photocatalytic CO2 reduction under solar illumination ascompared with previous studies (Table S2). A small amount ofethane (C2H6) was also detected among the Pt photodepositedblue titania samples whose results have been reported inTable S3, along with the selectivity for each photocatalyst stud-ied herein this work.
In comparison to 0.35-BT-30, a decrease in the CH4 evolutionand AQY was observed for 0.25-BT-30, 0.30-BT-30 and 0.32-BT-30. We suggest that the low Pt content did not provide sufficientelectron trap centers to hinder charge recombination, resultingin a reduced photoactivity. Higher Pt content beyond our opti-mum sample also manifests a decreasing trend in CH4 yield.Fig. S11 and Table 1 show, respectively, the size distributionand mean particle size of the Pt nanoparticles. Pt nanoparticlesize increases with increasing Pt precursor concentration, varyingfrom 1.82 nm for 0.25 Pt wt. %, to 3.5 nm for 0.50 Pt wt. %. It isknown that energy levels shift upward in very small nanoparti-cles [13].
Catalyst stability is of utmost importance in realizing a practi-cal system. Fig. 6d shows the stability of representative sample
1.5G illumination for: (a) P25 and reduced titania samples, and (b) Y-BT-30ytic CO2 conversion for different Pt-sensitized BT-30. (d) Stability test of 0.35-a vacuum oven at 100 �C for 2 h.
RES
EARCH:Original
Resea
rch
Materials Today d Volume 20, Number 9 d November 2017 RESEARCH
0.35-BT-30 upon sequential testing. The decrease in productionrate during the second cycle (without thermo-vacuum treat-ment) suggests a gradual deactivation of the photocatalyst, aphenomenon commonly observed in CO2 photoreduction stud-ies [14,29] and believed due to saturation of adsorption sites onthe titania surface with intermediate products [14,29]. However,the photocatalyst recovers its activity after being heated in a vac-uum oven at 100 �C for 2 h after 2nd, 3rd and 4th run. We havealso conducted the stability tests with a separate fresh batch of0.35-BT-30 sample, and it also shows a similar methane yieldand sample stability, thus confirming the reproducibility inmethane yield (Fig. S12).
To help rule out possible involvement of carbon impurities orsurface contaminants, a photoconversion test was run using amixture of Ar(g) and H2O, and the reactor was illuminated inthe presence of moist CO2 but in the absence of photocatalyst.For both tests the measurable CH4 yield was negligible, seeFig. 7a.
A 13C-labeled isotopic experiment was performed, using13CO2 instead of standard 12CO2, with the resulting 13CH4 ana-lyzed by GC-MS, see Fig. 7b–d. Fig. 7b corresponds to 13CO2,while Fig. 7c shows the signal at m/z = 17 attributed to 13CH4
[15,30–32] confirming CO2 as the main carbon source for CH4
generation. We conducted prolonged photocatalytic isotopictests, Fig. 7d, finding an increase in 13CH4 from 1 h to 3 h. A sim-ilar observation in increase of isotopic gaseous products with pro-
FIGURE 7
(a) Control test results of sample 0.35-BT-30. GCMS spectrum of: (b) 13CO2 used13CH4 evolved during prolonged photocatalytic testing of 0.35-BT-30 under mo
longed illumination was reported previously [33]. Figs. S13–S15show optical properties and CO2 photoreduction results of vari-ous control samples, respectively.
Literature suggests the process of CO2 photoreduction is basedon proton coupled multi-electron transfer [13,15,34–41]. Workby Thampi et al. have elucidated the underlying mechanism togenerate CH4 from CO2 photoreduction, where electrons andholes created in the TiO2 bandgap play a key role in the metha-nation process [8]. To produce H+ ions the VBM should be at apotential more positive than that of H2O/O2 [34], while theCBM should be more negative than the CH4/CO2 potential. ForBT-30, UV–vis DRS analysis indicates a band gap of 2.73 eV. Fromthe VB-XPS results as shown in Fig. 8a, the VBM position for BT-30 is at 1.19 eV which is more positive than Eo(H2O/O2) (0.82 Vvs. NHE). The CBM is located to be at �1.54 eV which is morenegative than Eo(CH4/CO2) (�0.24 V vs. NHE).
Under illumination, photogenerated e�–h+ pairs are generatedin the CB and VB of the BT, (reaction 2); the electrons can trans-fer to the Pt nanoparticles due to the lower Fermi energy level[13]. The holes in the valence band can oxidize adsorbed H2Ointo O2 and protons (H+) because the VBM position is more pos-itive than that of Eo(H2O/O2) [15,20,30,34,41–46] (reaction 3).The formation of CH4 requires 8e� and 8H+. The enriched e�
density on the Pt nanoparticles would favor CH4 formation,which is thermodynamically more feasible than CO, by transferof e� from the CBM of BT-30 to Pt, and from there to surface
during the experiment; (c) 13CH4 produced from 0.35-BT-30 in 1 h; and (d)ist 13CO2 atmosphere.
513
FIGURE 8
(a) Proposed energy band diagram of P25 nanoparticles and sample BT-30, and (b) possible transfer of electrons in 0.35-BT-30 under AM 1.5G illumination forphotoreduction of CO2 into CH4.
RESEA
RCH:O
riginal
Research
RESEARCH Materials Today d Volume 20, Number 9 d November 2017
adsorbed CO2 molecules [18] (reaction 4). Fig. 8b illustrates a pos-sible mechanism by which CO2 could be photoconverted intoCH4, based upon our experimental results for 0.35-BT-30.
Pt-BTþ hm ! Pt-BTðTi3þe�CBÞ þ Pt-BTðVohþVBÞ ð2Þ
4H2Oþ Pt-BTðVo8hþVBÞ ! 8Hþ þ 2O2 ð3Þ
CO2 þ 8Hþ þ Pt-BTðTi3þ8e�CBÞ ! CH4 þ 2H2O ð4Þ
4. ConclusionsA rapid, low-temperature technique for reducing titania wasdeveloped; the H2 desorbed from NaBH4 create oxygen vacancieschanging the valence of Ti4+ to Ti3+. Physiochemical and opticalproperties of the blue titania were analyzed, and a photocatalystwith optimum defects, narrow band gap, well-aligned band posi-tion, and reduced charge recombination for photocatalytic CO2
reduction into CH4 was obtained. The loading of a small amountof Pt (0.35 wt. %) onto the blue titania samples was found to sig-nificantly enhance CH4 production, showing an increase of 29�over the pristine blue titania. As revealed by ICP-OES and HR-TEM analysis, it was found that both the size and quantity ofthe Pt nanoparticles play a key role in determining photocat-alytic efficiency. The maximum CH4 production rate obtainedwas 80.35 lmol g�1 h�1 under simulated solar light, a value supe-rior to those of previously reported photocatalyst materials. Afterundergoing desorption, the catalyst shows appreciable stable per-formance over five 6 h cycles, the limit of our testing. Severalcontrol tests and a 13C isotopic tracer experiment were con-ducted to confirm CO2 as the carbon source. The outstandingactivity and stability of our reduced titania samples suggest thata balanced combination of surface defects, oxygen vacancies,band position, and optimized amount of Pt cocatalyst enablesan efficient photocatalyst for photoconversion of CO2 to fuel.
514
AcknowledgmentsThe authors gratefully acknowledge the support of the DGIST
R&D Program of the Ministry of Science and ICT (17-01-HRLA-01). This research was also supported by the Technology Devel-opment Program to Solve Climate Changes of the NationalResearch Foundation (NRF) funded by the Ministry of Scienceand ICT (NRF-2015M1A2A2074670).
Appendix A. Supplementary dataSupplementary data associated with this article can be found, inthe online version, at https://doi.org/10.1016/j.mattod.2017.09.005.
Appendix B. Apparent quantum yield (AQY)calculationApparent quantum yield (AQY) is defined as the ratio of numberof reacted electrons to the number of incident photons. The gen-eral equation is given as below [13,47,48]
AQY ð%Þ ¼ Number of reacted electronsNumber of incident photons
� 100 ðA:1Þ
As given in equation (A.2), during the photocatalytic CO2
reduction, stoichiometrically 8 electrons are required to produceone molecule of CH4 [13]
CO2 þ 8Hþ þ 8e� ! CH4 þ 2H2O ðA:2ÞTherefore, number of reacted electrons can be calculated by
directly multiplying 8 with mole of CH4 produced during thephotocatalytic reaction [13]
Number of reacted electrons ¼ ½CH4� � 8�NA ðA:3Þ
where [CH4] represents mole of CH4 produced in time (t) and NA is theAvogadro’s number (6.022 � 1023 mol�1). Herein, we describe the cal-culation for 0.35-BT-30 sample, which yield 80.35 mmol g�1 h�1 CH4.
RES
EARCH:Original
Resea
rch
Materials Today d Volume 20, Number 9 d November 2017 RESEARCH
Therefore,
number of reacted electrons ¼ 3:87� 1020 ðA:4ÞThe number of incident photons is same as the photon flux density[49], which is given by Eq. (A.5)
Number of incident photons ¼ Light absorbed by the photocatalystThe average photon energy
� t
ðA:5Þwhere “t” is the photoreaction time (s). Thus, light absorbed by thephotocatalyst is given by:
Light absorbed by the photocatalyst ¼H �A¼0:49Js�1
ðA:6ÞH is the apparent light input at the photocatalyst (Wm�2) of
100 W xenon solar simulator (Oriel, LCS-100) with an AM 1.5 fil-ter used in the present study (H = 1000 Wm�2) [13,50,51]. Simi-lar assessments were reported by Grätzel et al. [52]. Arepresents the geometric irradiation area of the photocatalystplaced inside the photoreactor, which is 0.00049 m2.
The average photon energy is given by ¼hck
¼ 5:64�10�19 J
ðA:7Þwhere h is Planck’s constant (h = 6.626 � 10�34 Js), c = 3 � 108 ms�1
and the value of k is found as average wavelength of the broadbandlight source (352 nm). Since the bandgap of BT-30 as estimated fromTauc plot analysis is 2.73 eV, therefore, unbound and free electron-hole pairs cannot be generated for k >454 nm. Thus, photons withwavelength 250–454 nm can excite electrons, whose average pho-tonic wavelength (k = 352 nm) is used to calculate the apparent quan-tum efficiency [51,53].Combining values from Eqs. (A.6) and (A.7)and substituting in Eq. (A.5)
Number of incident photons ¼ 3:12� 1021 ðA:8ÞSubstituting values from (A.4) and (A.8) in Eq. (A.1), the AQY
of sample 0.35-BT-30 is found to be 12.40%. The AQY values forall samples are given in Table 1.
References
[1] M.I. Hoffert et al., Science 298 (2002) 981–987.[2] K. Kim et al., RSC Adv. 6 (2016) 38964–38971.[3] D.C. Grills, E. Fujita, J. Phys. Chem. Lett. 1 (2010) 2709–2718.[4] L. Liu, C. Zhao, Y. Li, J. Phys. Chem. C 116 (2012) 7904–7912.[5] S.C. Yan et al., Angew. Chem. Int. Ed. 122 (2010) 6544–6548.
[6] V.P. Indrakanti, J.D. Kubicki, H.H. Schobert, Energy Environ. Sci. 2 (2009) 745–758.
[7] S. In et al., J. Am. Chem. Soc. 129 (2007) 13790–13791.[8] R. Thampi, J. Kiwi, M. Graetzel, Nature 327 (1987) 506–508.[9] R. Thampi, L. Lucarelli, J. Kiwi, Langmuir 7 (1991) 2642–2648.[10] S.Y. Zhu et al., Phys. Chem. Chem. Phys. 17 (2015) 9761–9770.[11] A. Sinhamahapatra, J. Jeon, J. Yu, Energy Environ. Sci. 8 (2015) 3539–3544.[12] J. Zhao et al., Appl. Catal. A: Gen. 510 (2016) 34–41.[13] W.N. Wang et al., J. Am. Chem. Soc. 134 (2012) 11276–11281.[14] W.J. Ong et al., RSC Adv. 3 (2013) 4505–4509.[15] A. Razzaq et al., Appl. Catal. B: Environ. 215 (2017) 28–35.[16] M. Li et al., Nano Energy 19 (2016) 145–155.[17] S. Docao et al., Energy Environ. Sci. 10 (2017) 628–640.[18] Y. Wang et al., RSC Adv. 4 (2014) 44442–44451.[19] T.A. Kandiel, A.A. Ismail, D.W. Bahnemann, Phys. Chem. Chem. Phys. 13
(2011) 20155–20161.[20] S.-I. In, D.D. Vaughn, R.E. Schaak, Angew. Chem. Int. Ed. 51 (2012) 3915–3918.[21] B. Banerjee et al., Nanoscale 7 (2015) 10504–10512.[22] D.S. Kim, J.D. Jeon, K.H. Shin, Microporous Mesoporous Mater. 181 (2013) 61–
67.[23] Z. Wang et al., Adv. Funct. Mater. 23 (2013) 5444–5450.[24] R. Fu et al., RSC Adv. 4 (2014) 37061–37069.[25] Y. Shan et al., RSC Adv. 5 (2015) 34737–34743.[26] M. Senna et al., J. Solid State Chem. 187 (2012) 51–57.[27] Y. Yan et al., J. Mater. Chem. A 2 (2014) 12708–12716.[28] W. Ong et al., Dalton Trans. 44 (2015) 1249–1257.[29] L. Liu et al., Appl. Catal. B: Environ. 134–135 (2013) 349–358.[30] T. Yui et al., ACS Appl. Mater. Interfaces 2 (2011) 2594–2600.[31] S. Zhu et al., Green Chem. 18 (2016) 1355–1363.[32] X. Meng et al., Chem. Commun. 50 (2014) 11517–11519.[33] K. Li et al., Nanoscale 6 (2014) 9767–9773.[34] S.N. Habisreutinger, L. Schmidt-Mende, J.K. Stolarczyk, Angew. Chem. Int. Ed.
52 (2013) 7372–7408.[35] S.M. Park et al., ACS Omega 1 (2016) 868–875.[36] K. Mori, H. Yamashita, M. Anpo, RSC Adv. 2 (2012) 3165–3172.[37] B. Fang et al., ACS Sustain. Chem. Eng. 3 (2015) 2381–2388.[38] H. Liu et al., Langmuir 32 (2015) 254–264.[39] J.C. Wang et al., ACS Appl. Mater. Interfaces 8 (2016) 3765–3775.[40] L. Yu, ACS Catal. 6 (2016) 6444–6454.[41] L. Tan et al., Appl. Catal. B: Environ. 166–167 (2015) 251–259.[42] M. Li et al., Appl. Catal. B: Environ. 201 (2017) 629–635.[43] S. Zhu et al., Appl. Catal. B: Environ. 187 (2016) 11–18.[44] J. Hou et al., Nano Energy 30 (2016) 59–68.[45] G. Yang et al., Appl. Catal. B: Environ. 219 (2017) 611–618.[46] X. Liu et al., Sci. Rep. 6 (2016) 38474.[47] L. Tan et al., Chem. Eng. J. 308 (2017) 248–255.[48] W.J. Ong et al., Nano Res. 7 (2014) 1528–1547.[49] Y. Ding, P. Nagpal, Nanoscale 8 (2016) 17496–17505.[50] H. Rim et al., J. CO2 Util. 20 (2017) 91–96.[51] M. Zubair et al., J. CO2 Util. 20 (2017) 301–311.[52] M. Schreier et al., Nat. Energy 2 (2017) 17087.[53] W. Ong et al., Nano Energy 13 (2015) 757–770.
515